trybot +1 doesn't trigger a new tryjob build |
||
Issue descriptionTake an example of CL https://chromium-review.googlesource.com/c/chromiumos/containers/cros-container-guest-tools/+/1403876 It kicks off pre-cq at about 1:30, at that time a bad CL is in ToT, which will cause this round of pre-cq fails. Revert for the bad CL (https://chromium-review.googlesource.com/c/chromiumos/chromite/+/1416430) is in at 13:57. 2:57, the first failed tryjob appears. After that, the owner set trybot +1 again, a new round of pre-cq is kicked off at 4:03. At 4:24, a pre-cq failure comes again: https://cros-goldeneye.corp.google.com/chromeos/healthmonitoring/buildDetails?buildbucketId=8924140167090159728 However, this failed tryjob is not in the round of 4:03 pre-cq, it's kicked off at 13:31, so it's in the 1:30 pre-cq. It's very confusing that another trybot+1 won't trigger a new tryjob, but inherit the last failed one. CQ+1 or trybot+1 both can't make this CL submitted and get tested correctly.
,
Jan 17
(6 days ago)
I agree that it's confusing. It happened because a doomed build was in progress when the +1 was re-added. The system doesn't retrigger in-progress builds to avoid the waste. However, the fact that a PreCQ build took 3 hours seems like a big deal and substantially exaggerates the problem..
,
Jan 17
(6 days ago)
It's also extra confusing because you get a failure message which doesn't indicate that there's more failures in progress, and give a way to stop them immediately instead of continuing them. Also, there's no useful feedback to indicate what's happening which further confounds the issue because it involves roping in other people to find out what is going on. I thought that the failures at 4:34 were the ones started at 4:03pm.
,
Jan 17
(6 days ago)
The Parallel CQ design will address this, partly by displaying builder state directly in Gerrit. Further the PreCQ will be deprecated as that goes live.
,
Jan 17
(5 days ago)
We'll completely delete all the PreCQ code when PCQ launches and this behavior has been here for years. Thanks for the report and sorry for the confusing existing behavior. |
||
►
Sign in to add a comment |
||
Comment 1 by xixuan@chromium.org
, Jan 17 (6 days ago)